How much to spend on creatives a month: the media-tier method

Allow 15 to 30% of your monthly media budget for a testing pot, and work your creative budget out from it. Divide that pot by the read cost of a creative, roughly 2.5 to 3 times your target CPA, and you get the number of creatives you can genuinely judge this month. In practice, an account with €10,000 of media reads 15 to 20 creatives a month, which means €2,000 to €4,000 of production. Producing more means buying videos the algorithm will never look at. A creative budget is not a percentage: it is a division. According to the 35e Observatoire de l'e-pub (35th e-advertising Observatory) from the SRI and UDECAM, the French digital advertising market reached €12.4bn in 2025, up 11%, driven notably by Social and Video.

How much should you actually spend on creatives each month?

Your monthly creative budget is calculated backwards: it is deduced from the number of creatives your media budget can read in the month, not from a percentage applied to your revenue. Every video produced beyond that reading capacity is a video you pay for and that nobody, neither you nor the algorithm, will ever judge. It will sit in drafts, or run for three days with €40 behind it before being cut “because it doesn't work”.

The calculation chain fits on four lines:

• Monthly media budget × testing share (15 to 30%) = testing pot
• Testing pot ÷ read cost of one creative (≈ 2.5 to 3 × target CPA) = number of readable creatives
• Readable creatives × average unit cost (concept/iteration mix) = monthly creative budget in euros
• Creative budget ÷ media budget = production/media ratio, the figure you will present to your leadership

A concrete case, the one we see most often with our French e-commerce clients. A Meta account with €10,000 of media a month, target CPA €40. Testing pot at 20%: €2,000. Read cost: €110 per creative. Result: 18 readable creatives over the month. At €150–200 each in a mix of concepts and variants, the creative budget lands between €2,700 and €3,600. Ruche & Pollen arrived at the same order of magnitude in 2025 on the French side: an account at €10,000/month should run with 15 to 20 active creatives. Two methods, two sources, the same figure. That is reassuring.

Below €1,500 of media a month, the calculation gives 2 readable creatives. Two. At that level, it is no longer a creative budget to arbitrate, it is the media floor below which video is pointless. We come back to it at the end of the article: it is the most common case and the least well handled.

Second half of the calculation: what it costs in euros of production, and what that means as a weekly pace.

Three reading notes, and they matter. The “new creatives / week” columns come from the Motion dataset Creative Benchmarks 2026 (578,750 creatives, 6,015 accounts, $1.29bn of Meta spend between September 2025 and early January 2026), published in 2026 and picked up by Conduit SA. The original tiers are in dollars of monthly spend; we have aligned them to nearby euro tiers, which is not an exchange-rate conversion. The read cost (2.5 to 3 × target CPA) is our own sizing rule, cross-checked against Spark UGC's $120 “read budget” in 2025 and the £300–500 per test cited by Adamigo the same year. Finally, the production/media ratio falls as the budget rises: counter-intuitive, and explained further down by the concept/iteration mix.

Why is “10% of media on production” a poor benchmark?

Because that benchmark comes from television, where targeting was bought in GRPs and the film was a fixed cost amortised over a year. Simulmedia was still citing the old 20/80 rule in 2024: 20% production, 80% airtime. One spot, six months on air, a single execution. In paid social, the creative is the targeting. You no longer decide who sees what: you offer different inputs to a system that decides for you. Changing the creative means changing the audience. Meta detailed this in December 2024 when presenting its Andromeda engine: the first stage of its ad recommendation system keeps a few thousand relevant ads out of tens of millions of candidates.

A fixed percentage breaks for a mechanical reason: it ignores your CPA. Two brands at €10,000 of media, one with a target CPA of €15, the other at €80. The first can read 40 creatives in the month. The second will read 6. Same media budget, same “10%”, and yet a production gap of one to seven. A percentage cannot see that. A division can.

The serious rules found on the market never talk about production anyway; they talk about testing. The Social Outline champions “the 15% rule”: 15% of monthly spend allocated to unproven creative, with a minimum per derived concept based on cost per acquisition, never a flat amount. Spark UGC starts from an 80/20 split and pushes it to 30% when fatigue is confirmed, then brings it back down to 15% when one creative carries the account. On the French side, the DHS Digital newsletter cited a 20% testing budget benchmark in 2025, with a multiple of 2 to 4 applied to the CPA to size the volume. These three sources say the same thing in different words: the variable is the testing pot. The production budget is its consequence.

A production percentage tells you how much to spend. A division tells you how many decisions you will be able to make this month. Only the second piece of information is any use in a budget meeting.

How do you calculate the read cost of a creative in your account?

Read cost is the minimum amount to spend behind a creative to earn the right to say anything about it. We set it at 2.5 to 3 times the target CPA, and it is a sizing rule, not statistical significance. Better to be blunt about this, because a lot of content sells the opposite.

What you buy at 2.5–3 × CPA: enough impressions to read a hook rate, a hold rate at 3 and 6 seconds, an outbound CTR, sometimes an add-to-cart rate. Enough to decide between “we iterate” and “we bin it”. What you do not buy at that price: proof of CPA. Novoads reminded readers in late 2025 that a read needs around 100 events to mean anything, and that an account's real ceiling is the events available, not the number of videos. If your reference event is a €40 purchase, 100 purchases represent €4,000 behind a single creative. Nobody tests like that.

Buyers know the way out of this trap: you read creatives on a high-volume upstream signal (3-second view, click, ATC) and only demand proof of CPA from the creatives promoted into a scaling campaign. Meta, for its part, wants 50 conversions per week per ad set to exit learning, and Ruche & Pollen put figures in 2025 on what that implies in practice: at a €20 CPA, €1,000/week per ad set, so about €4,000/month for a single set, and €15,000 to €20,000/month for a mature account with three or four sets. Your read cost and your campaign structure constrain each other.

The most telling example comes from Spark UGC, in 2025: a $1,000 testing pot divided by a $120 read budget gives 8 tests a month, so two a week. The same advertiser who decides to launch 15 creatives on the same pot drops to $66 per creative, below the reading floor. They did not test twice as much. They stopped testing.

What we do not know, and must own up to: the right multiple varies with the volatility of the account. With a low average basket and a high purchase volume, 2 × the CPA is often enough. For long-cycle B2B or a €300 product, we go up to 4 × and read on the qualified lead. Take the middle multiple, measure for three months, adjust. The figure is not sacred; the discipline of never dropping below the floor is.

How many new creatives per week at your spend tier?

Between 2 and 5 new creatives per week under $50,000 of monthly spend, 7 to 16 between $50,000 and $200,000, and over 30 for the top quartile of accounts above a million. These are the figures from the 2026 Motion dataset, picked up by Conduit SA: an average of 2.8 new creatives/week under $10,000, 4.1 between $10,000 and $50,000, 6.7 between $50,000 and $200,000, 11.2 between $200,000 and $1m, 18.9 above. The top 25% of each tier produce roughly double: 4.8 / 8.1 / 16.0 / 31.1 / 54.6.

The column that hurts is the winners one. Again according to Motion, only 5 to 8% of ads become true winning creatives, and nearly half never receive meaningful spend. In winners per month: Motion measures about 0 winners/month under $10,000, 0.25 between $10,000 and $50,000, 0.75 between $50,000 and $200,000, 1.75 between $200,000 and $1m, and several a month beyond a million. An account with $30,000 of monthly media produces one winning creative per quarter. That is not a failure, it is the market norm.

Other sources converge with a different breakdown. Roaspig gave in 2025: 5 to 10 tests/week between $5k and $15k/month, 10 to 25 between $15k and $50k, 25 to 50 between $50k and $150k. MHI Growth Engine, on an analysis of 80 DTC accounts in 2025, offers a simpler benchmark to remember: one new concept for every $2,000 to $3,000 of monthly spend. Foxwell Digital, in 2024, to be taken as a historical benchmark, reasoned at $2,000 per test, i.e. 50 creatives for $100,000 of spend and 25 to 50% of the budget in testing at that level. Billo published a synthesis in early 2026 combining two reports representing $3.7bn of spend, broken down by tier, platform and refresh cadence; the orders of magnitude remain the same.

Beware the translation trap. These benchmarks are in dollars and mostly concern US accounts. A French advertiser at €5,000/month will not find their line anywhere in these reports, hence the value of going back to the division rather than copying a number of videos per week read on LinkedIn. And before committing to a pace, look at how long that pace takes to produce a result: most disappointments come from a judgement horizon set at three weeks.

Concepts or iterations: where does the monthly budget actually go?

The concept is expensive and rarely tested; the iteration is the real unit of purchase, and it is what makes a ceiling of 50 or 100 creatives a month sustainable. A monthly creative budget is steered by adjusting the ratio between the two, not by negotiating the unit price.

The cost ranges above are our own orders of magnitude for vertical-video production, not a published rate card. They shift with the product, the shoot location and the creator's profile. One external benchmark to calibrate: Aden's Lab found in 2025 that standard UGC ran between $50 and $300 per video, plus 30 to 50% in usage fees, and concluded that testing 30 to 50 creatives a month at that rate is out of reach for most advertisers.

These usage fees are the line item monthly budgets systematically forget. They are not production costs: they are rent. You pay them again at every rights renewal, including on videos that no longer run. On a creative budget of €3,000/month, a badly bounded cascade of rights can freeze €600 to €900 without producing a single new creative. The detail is here: the real rate card and the 5 hidden costs.

The mix also explains why the production/media ratio falls as the account grows. At €3,000 of media, nearly every creative is a new shoot: the average cost stays high. At €60,000, you have three or four winning concepts in stock and you make forty variants around them: new hooks, new opening shots, new text hooks, 9:16 and 4:5 recuts. The marginal cost collapses. A big account spends proportionally less on creation because it has creative capital to amortise, not because it negotiates better.

What happens if you produce too many creatives for your media budget?

You dilute your testing pot, drop below the read floor, and make decisions on noise. That is the over-production ceiling, and nobody talks about it because the whole market, ourselves included, since we sell production, has an interest in pushing volume.

The mechanism is simple. The pot is constant. Each extra creative takes a share away from the others. At 15 creatives for a €1,000 pot, each gets €66: too little to separate a genuinely good hook from a chance variation over 400 impressions. You cut a creative that would have won, you keep one that got lucky, and you start again the next month convinced that “nothing works”. Fifteen tests at €66 are not worth eight tests at €120. They are worth less, because they produce false conclusions that you then feed back into the next briefs.

Second symptom: dormant stock. Brands come to us with 60 videos delivered over the quarter and 11 put live. The other 49 cost exactly their purchase price for a value of zero. From the finance department's point of view, that is a creative budget that has doubled with no effect on CPA. From the media buyer's point of view, it is a library they never had the means to read.

We often hear the objection of platform limits. Meta does cap the number of ads per Page based on the highest month of spend: 250 ads under $100,000, 1,000 under $1m, 5,000 under $10m, 20,000 beyond (documentation relayed by Novoads). In real life, no French advertiser at €10,000/month gets near 250 active ads. The real ceiling is economic, not technical: it is the number of conversion events your spend generates. The trade site PPC Land reported in September 2026, incidentally, that some Meta advertisers now see a message in Ads Manager saying the number of ads per Page is no longer limited.

Last point, and the most common: an urgent volume order is very often a wrong diagnosis. Performance drops, creative fatigue is declared, production is tripled. Before approving that budget line, check that it really is fatigue and not a seasonal CPM increase, an attribution change or a product page problem. Adamigo noted in 2025 a CPM rise of around 18% at the start of the year in competitive sectors such as fashion and beauty. A CPM that rises 18% degrades your metrics without a single one of your creatives having aged.

How do you evolve this creative budget month after month without breaking everything?

You only move the creative budget after a media tier change that has held two months in a row, never in reaction to a single week. The creative budget is a subscription indexed to spend, not a mood variable.

The rule we apply on our accounts, in three moves:

• Raise when media has crossed into the next tier for two consecutive months and the share of creatives actually put live exceeds 80% of deliveries. If you are not running what you already have, producing more solves nothing.
• Freeze when one creative carries more than 40% of spend. Spark UGC recommends dropping testing back to 15% in that case: you protect the cash cow, iterate around it, and stop costly new concepts.
• Push to 30% testing only on confirmed fatigue: frequency up, hook rate down on the three main creatives, CPM stable. All three at once, otherwise it is something else.

Seasonality deserves a line of its own, because it breaks every linear annual budget. In November, your CPM goes up, so your read cost goes up, so with a constant testing pot you read fewer creatives. Two options: raise the pot in proportion to CPM, or accept testing less and enter Q4 with a stock of creatives already validated in September and October. The second is almost always the right one. You build the library when impressions are cheap, and run it when they are expensive.

A word on the direction of the market, because it justifies budgeting creative as a recurring line. The IAB put US creator ad spend in November 2025 at a projected $37bn for the year, growing roughly four times faster than the overall media market, based on a survey of more than 450 budget decision-makers. Motion noted for its part in 2025 that 42% of the best-funded ads were classed as “lo-fi”. Vertical production is no longer an exceptional item you free up once a year for a campaign. It is a monthly expense, like the subscription to your emailing tool. The same IAB creator economy report indicated in 2025 that 48% of US advertisers surveyed consider creators a must-buy, behind only social networks and paid search.

How much should you invest in creatives per month when you spend under €3,000 on media?

Under €3,000 of monthly media, the right answer is often: do not raise your creative budget yet, put the money into media until you reach the floor. It is the recommendation that costs us contracts, and the one we give anyway.

French benchmarks locate the zone well. The Observatoire de la publicité gave in late 2025: €500 to €1,000/month for a start-up, €3,000 to €10,000/month for an established SME, several tens of thousands for large accounts. Vision Plus cited around the same time a realistic range of €300 to €800/month of pure media for a French SME in 2026. Eidos Agency, in 2024, spoke of €20 to €50/day for a beginner, i.e. €600 to €1,500/month. A large part of the French e-commerce landscape lives in this band. And in this band, the calculation does not give a creative budget: it gives two readable creatives a month.

What we concretely advise an advertiser at €2,000 of media:

• Three to four solid concepts at launch, €800 to €1,200 in one go, then nothing more for two months.
• A testing pot at 15%, i.e. €300, which funds two clean reads a month, not five dirty ones.
• Iterations only, at €300–500/month, until media has crossed €3,000.
• Zero renewals of usage rights on creatives that no longer run. You cut, and only renegotiate on the winners.

A food-supplements brand we work with stayed eleven months at €2,500 of media with four concepts and about ten iterations in total over the year. It did not lack creativity: it lacked the budget to judge. The day media went up to €9,000, the same concepts could finally be read, and two of them had already been carrying performance for months without our being able to prove it.

The reflex we would like to see disappear in 2026: setting your creative budget by looking at what a competitor does on LinkedIn. Your useful ceiling is written in your ad account, not in theirs. Open Ads Manager, take last month's spend and your target CPA, apply the division. If the result seems ridiculously low, the problem is not production: it is that you have too little media to have the right to an opinion about your videos.

Frequently asked questions

What percentage of my media budget should go on creatives?

Between 10 and 35% depending on your tier, but that percentage is an outcome, not a rule. It is derived from the testing pot (15 to 30% of media) divided by the read cost of a creative. Small accounts naturally climb to 30%; accounts above €100,000 of media drop to 10–18% thanks to iterations.

How many creatives do you need for a €10,000 monthly media budget?

Fifteen to twenty active creatives over the month, a figure put forward by Ruche & Pollen in 2025 and consistent with our calculation: €2,000 of testing pot divided by €110 of read cost gives 18 judgeable creatives. That is a production budget of €2,500 to €3,500 in a mix of concepts and iterations.

What is the read cost of a creative?

The minimum amount to spend behind an ad to earn the right to draw a conclusion from it. We set it at 2.5 to 3 times the target CPA. Below that, the differences observed are down to chance. Spark UGC cited a read budget of $120 per test in 2025; Adamigo spoke of £300 to £500 to reach significance in the UK.

Can you produce too many creatives per month?

Yes, and it is common. Beyond the reading capacity of your testing pot, each extra video reduces the budget allocated to the others and takes you below the statistical floor. The result: false conclusions and a stock of videos never put live. Fifteen tests at €66 are worth less than eight tests at €120.

How many winning creatives can you expect per month?

According to the Motion Creative Benchmarks 2026 dataset, only 5 to 8% of ads become true winners. In volume: about zero a month under $10,000 of spend, 0.25 between $10,000 and $50,000, 0.75 between $50,000 and $200,000, and 4 above a million. An average account produces one winner per quarter.

Should you raise the creative budget when performance drops?

Not before confirming creative fatigue: frequency rising, hook rate falling on the main creatives and CPM stable. If CPM climbs, and Adamigo noted +18% in early 2025 in fashion and beauty, your metrics degrade without your videos having aged. Producing more will fix nothing.